Appl. Phys. Lett. 115, 083503 (2019); https://doi.org/10.1063/1.5116811 115, 083503

© 2019 Author(s).

Electromagnetic induction imaging withatomic magnetometers: Unlocking the low-conductivity regimeCite as: Appl. Phys. Lett. 115, 083503 (2019); https://doi.org/10.1063/1.5116811Submitted: 27 June 2019 . Accepted: 06 August 2019 . Published Online: 21 August 2019

Luca Marmugi , Cameron Deans , and Ferruccio Renzoni

ARTICLES YOU MAY BE INTERESTED IN

The effect of luminescent coupling on modulated photocurrent measurements inmultijunction solar cellsApplied Physics Letters 115, 083506 (2019); https://doi.org/10.1063/1.5115014

Current-injection quantum-entangled-pair emitter using droplet epitaxial quantum dots onGaAs(111)AApplied Physics Letters 115, 083106 (2019); https://doi.org/10.1063/1.5103217

High-responsivity photodetectors made of graphene nanowalls grown on SiApplied Physics Letters 115, 081101 (2019); https://doi.org/10.1063/1.5097313

Electromagnetic induction imaging with atomicmagnetometers: Unlocking the low-conductivityregime

Cite as: Appl. Phys. Lett. 115, 083503 (2019); doi: 10.1063/1.5116811Submitted: 27 June 2019 . Accepted: 6 August 2019 .Published Online: 21 August 2019

Luca Marmugi,a) Cameron Deans, and Ferruccio Renzoni

AFFILIATIONS

Department of Physics and Astronomy, University College London, Gower Street, London WC1E 6BT, United Kingdom

a)Electronic mail: l.marmugi@ucl.ac.uk

ABSTRACT

Electromagnetic induction imaging with atomic magnetometers has disclosed unprecedented domains for imaging, from security screeningto material characterization. However, applications to low-conductivity specimens—most notably for biomedical imaging—require sensitiv-ity, stability, and tunability only speculated thus far. Here, we demonstrate contactless and noninvasive imaging down to 50 S m�1 using a50 fT/

ffiffiffiffiffiffi

Hzp

87Rb radio frequency atomic magnetometer operating in an unshielded environment and near room temperature. Two-dimensional images of test objects are obtained with a near-resonant imaging approach, which reduces the phase noise by a factor 172, witha projected sensitivity of 1 S m�1. Our results, an improvement of more than three orders of magnitude on previous imaging demonstrations,push electromagnetic imaging with atomic magnetometers to regions of interest for semiconductors, insulators, and biological tissues.

Published under license by AIP Publishing. https://doi.org/10.1063/1.5116811

Imaging is ubiquitous throughout science, technology, and every-day life. However, some areas remain not directly accessible. This isbecause of the lack of dedicated instrumentation, or the limited sensi-tivity of established technologies. This is the case of conductivity map-ping of biological tissues and organs, which is precluded to thecurrently deployed diagnostic devices. Contactless and noninvasiveaccess to such information would provide insight and diagnostic toolsfor several conditions, whose clinical phenotype features modificationsof the conductivity of the involved tissues. Examples include degenera-tive neurological and muscular diseases, fibrosis, skin healing, liver,lung, and mammary gland tumors,1 hemorrhages, edemas, and car-diac fibrillations. Conditions such as atrial fibrillation have enormoushuman and economic costs due to nonideal diagnostics and limitedaccess to the fundamental pathogenic mechanisms.2

A possible solution to the problem of identifying suitable imagingdevices is magnetic induction tomography,3 which has been proposedfor many applications in the biomedical field.4–7 In this approach, anoscillating magnetic field induces eddy currents, whose densitydepends on the local properties of the specimens, and a magnetometermeasures the secondary field produced by them. However, the limitedsensitivity and tunability of the magnetic sensors in use—most oftenpickup coils—have prevented suitable imaging performance.Detection of low-conductivity objects (�1 Sm�1) has only been

achieved with conventional sensors for very large volumes (e.g.,�2000 cm3).8

Electromagnetic induction imaging with radio frequency (RF)atomic magnetometers9,10 has the potential to overcome such limita-tions. It has been successfully demonstrated in a number of applica-tions, including localization and identification of nonmetallic targets,11

material characterization,11–14 and security screening.15,16 Thus far,imaging has been limited to large conductivities (�105 S m�1).Recently, high-conductivity imaging has been also obtained withnitrogen-vacancy (NV) center magnetometers,17 where the proximityto the specimen reachable with a nanosized sensor can compensate forthe current lack of sensitivity. However, none of these demonstrationshas fully met the requirements for practical uses in conductivity map-ping of biological tissues. Namely: (i) high sensitivity for measuringthe small response of nonconductive specimens; (ii) broad tunabilityfor controlling the penetration and compensating for the decrease inconductivity; (iii) high-phase stability in unshielded environments forensuring reliable imaging; and (iv) large dynamic range (i.e., the abilityto simultaneously image large differences in conductivity) to imagespecimens with large variations in conductivity, without any detrimen-tal saturation effect.

In this letter, we address the above points, demonstrating 2Delectromagnetic induction imaging of 6.25 cm3 specimens down to

Appl. Phys. Lett. 115, 083503 (2019); doi: 10.1063/1.5116811 115, 083503-1

Published under license by AIP Publishing

Applied Physics Letters ARTICLE scitation.org/journal/apl

50 S m�1. Images are obtained with a 50 fT/ffiffiffiffiffiffi

Hzp

, broadly tunable87Rb radio frequency (RF) atomic magnetometer,18–21 operating in amagnetically noisy environment. Specimens are imaged in real time,in the form of a phase map, directly proportional to the local conduc-tivity. Phase stability better than 0.03� across the imaging area overseveral hours is obtained by biasing the RF atomic magnetometer tonear-resonant operation. In this way, the impact of residual magneticnoise is reduced by two orders of magnitude. Simultaneous imaging ofsamples spanning three decades in conductivities is also demonstrated.Our results represent a 1460-fold improvement in conductivity overprevious studies performing electromagnetic imaging with atomicmagnetometers11—extending its domain to low-conductivity speci-mens. This moves applications to biomedical imaging closer.2

A sketch of the setup is shown in Fig. 1(a). The core of the systemis a cubic quartz chamber containing isotopically enriched 87Rb and40Torr of N2 as the buffer gas. The cube has a side of 25mm. Thevapor cell is housed in a 3D printed Nylon case, hosting an active tem-perature control system (4.6W capacity). The vapor is maintained at45 �C. The sensor is held in place by a polylactide (PLA) mount,secured to a marble table equipped with Sorbothane vibrationalisolators.

A three-axes Helmholtz and a three-axes quadrupole coil setsensure the homogeneity of the magnetic field at the sensor’s positionby canceling stray fields and gradients, respectively. A DC bias field(Bbias, along z) is actively stabilized by a proportional-integrative-derivative (PID) loop referenced to a fluxgate in close proximity to theatomic magnetometer. The fluxgate continuously monitors the back-ground magnetic field. Its output is used as the error signal of a feed-back loop.13,22 The magnetic field compensation system has aresponse band between DC and 1 kHz, imposed by the fluxgate.

Atomic spins are optically pumped by a circularly polarized laserdetuned by þ80MHz with respect to the D1 line ð52S1=2jF ¼ 1i! 52P1=2jF0 ¼ 2iÞ, at 795 nm. Tuning of the magnetometer isobtained with Bbias, collinear to a pump beam alongþz [see Fig. 1(a)].Optical pumping accumulates atoms in the jF ¼ 2;mF ¼ þ2i state.BRF (generated by a single ferrite-core coil of 7.8mm diameter, andoscillating along 6y) coherently drives atomic Zeeman coherences atxRF. Precessing spins are probed by a second laser, linearly polarized

and tuned to the D2 line (1.35 GHz to the blue side of the 52S1=2jF ¼ 2i ! 52P3=2jF0 ¼ 3i transition) at 780 nm [Fig. 1(b)]. The probebeam propagates along þx . Detection of Larmor precession isobtained through Faraday rotation of the probe beam’s polarizationplane. The rotation is measured by a polarimeter connected to a lock-in amplifier, referenced to xRF. Throughout this work, the amplifier’stime constant TC is set to 500ms. Operation of the RF atomic magne-tometer in the range between 100 kHz and 2MHz—the upper limitcurrently imposed by the acquisition electronics—is obtained.

Unlike conventional atomic magnetometry, which relies on themeasurement of spontaneously generated magnetic signals,23–26 in ourapproach, BRF(t, y) induces eddy currents in the samples, with a den-sity exponentially decaying along y [Fig. 1(c)] as determined by theskin effect,27 and a magnetic signature proportional to x2

RF , as alreadyreported.8 Specimens are placed in between the RF coil and the atomicmagnetometer, 30mm above it. However, this arrangement is not arequirement for electromagnetic induction imaging with atomic mag-netometers.12–14 Samples are supported by a motorized PLA andNylon support (step size 2mm). Eddy currents produce a secondaryfield (BEC) opposing BRF and phase-lagged with respect to it. BECcontains information on the conductivity r, permittivity e, perme-ability l, and the geometry of the samples.28,29 Such information isretrieved by monitoring the influence of BEC on the atomic spins’motion. Data are stored in position-referenced matrices. Each pixelis the result of a single measurement. Datasets are then smoothedwith a nearest-neighbor Gaussian filter (radius 4mm), and plottedin color-coded 2D graphs. The color map is scaled between themaximum and the minimum of each single image, unless otherwiseindicated. In the following, the symbol DU is used to convey theidea of a variation from the sensor’s equilibrium (i.e., no eddy cur-rents) induced by the sample and the related BEC. However, nobackground subtraction is performed.

Figures 2(a)–2(f) show phase maps (DU), obtained at 2MHz,spanning conductivities from r ¼ 1� 104 Sm�1 to 5� 101 Sm�1.This frequency was chosen as the upper limit of xRF, imposed by theacquisition electronics. The samples are 25� 25� 10mm3 Si blockswith decreasing n-doping concentrations (P, from 5� 1018 cm�3 to2� 1015 cm�3). Phase maps are presented as they are directly related

FIG. 1. (a) Sketch—to scale—of the 50 fT/ffiffiffiffiffiffi

Hzp

RF atomic magnetometer for electromagnetic induction imaging of low-conductivity specimens (laser systems not shown).The red and green lines mark the optical paths of the pump and the probe laser beams, respectively. The dark arrows indicate the orientation of the bias field (Bbias) and of thedriving field (BRF). (b)

87Rb level diagram and tuning of lasers. (c) Principle of the imaging technique. LIA, lock-in amplifier; SC, semiconductor sample; EC, eddy currents. Thedetails are in the main text.

Applied Physics Letters ARTICLE scitation.org/journal/apl

Appl. Phys. Lett. 115, 083503 (2019); doi: 10.1063/1.5116811 115, 083503-2

Published under license by AIP Publishing

to r. Furthermore, variations in phase cannot be produced by shield-ing or shadowing of the sensor.

In all cases, samples are clearly imaged with a satisfactory defini-tion. The 5� 101 Sm�1 sample surpasses the previous record by 1460times.11 The maximum recorded DU decreases with the samples’ con-ductivity and hence their dopant levels – with an approximately linearrelationship between DU and xRFr. Correspondingly, the phase varia-tion at a given frequency becomes smaller, as r decreases, unless xRF

is suitably increased. At 2MHz the skin depth dðxRFÞ is smaller thanthe sample thickness for the most conductive specimens. This reducedpenetration into the bulk explains why the maximum DU for the1� 104 Sm�1 sample [Fig. 2(a)] is smaller than that of the5� 103Sm�1 one [Fig. 2(b)].

Imaging single specimens does not provide any indication of theresponse in the presence of abrupt and large variations in conductivity.This is another essential requirement for biomedical imaging and navi-gation: for example, cancerous tissues can exhibit increases in conduc-tivity up to 500 times.1 To demonstrate the large dynamic range and theconductivity sensitivity of our system, a simultaneous image at 2MHzof the six samples of different conductivities is shown in Fig. 2(g). Thiscovers a change in conductivity of a factor 200. The low-conductivitysamples exhibit a decreasing contrast but are clearly visible—with signallevels consistent with the previous images.

The images shown in Figs. 2(a)–2(g) were only possible afterreducing the noise-induced phase variations. These hamper the imag-ing process: for imaging low-conductivity, phase stability is as criticalas the sensitivity and the tunability of the RF atomic magnetometer.Residual magnetic noise shifts the magnetometer frequency and per-turbs the recorded phase, imposing a limit on the smallest detectablephase change and therefore on the imaging performance. Magneticnoise is greatly reduced by operating within magnetically shielded envi-ronments, however this is not suitable for practical uses of electromag-netically induction imaging. Alternatively, self-tuning optimizationcould significantly improve the performance of active compensationsystems for atomic magnetometers.30

Instead, we introduce near-resonant induction imaging with RFatomic magnetometers, as illustrated in Fig. 3. Such a solution can be

FIG. 2. Imaging of low-conductivity samples with an RF atomic magnetometer operating at 2 MHz. DU scans of the doped Si specimens. (a) r ¼ 1� 104 Sm�1. (b)r ¼ 5� 103 Sm�1. (c) r ¼ 1� 103 Sm�1. (d) r ¼ 5� 102 Sm�1. (e) r ¼ 1� 102 Sm�1. (f) r ¼ 5� 101 Sm�1. (g) Simultaneous imaging at 2 MHz of specimens withthe conductivity spanning from ½1� 104 Sm�1; 5� 101 Sm�1�. Conductivity decreases from top left to right bottom. The samples are not in electrical contact.

FIG. 3. Near-resonant imaging. (a) Magnetometer amplitude response near 2 MHz.The nonlinear Zeeman effect causes the RF resonance to split in four components,corresponding to individual transitions jF ¼ 2;mFi $ jF ¼ 2;mF61i, as indi-cated by the respective labels. The shaded area highlights the region explored fornear-resonant operation. (b) Magnetometer phase profile and operating points fornear-resonant imaging. Inset: the entire phase response highlighting the near-resonant region. (c) Example of the measured phase noise (dUmeas) across theimaging area (the total phase noise in an image without specimens) vs detuning.(d) Calculated phase noise (dU5nT) produced by 5 nT magnetic field noise, as mea-sured across the imaging area. The circled numbers mark the position with respectto the resonance curve in panel (b). The vertical dotted lines mark the configurationused throughout this work.

Applied Physics Letters ARTICLE scitation.org/journal/apl

Appl. Phys. Lett. 115, 083503 (2019); doi: 10.1063/1.5116811 115, 083503-3

Published under license by AIP Publishing

extended to other magnetometers—independent of the performanceof the field stabilization, if applicable—and to any resonant, phase-sensitive system characterized by residual phase noise limiting itsperformance.

With this technique, the RF atomic magnetometer is biased tooperate in quasiresonant conditions [Fig. 3(a)], where the gradient ofthe phase response is the smallest [Fig. 3(b)]. This minimizes theimpact of magnetic field noise on the system’s output while maintain-ing the sensitive detection of phase changes DU arising from BEC. Toquantify the performance improvement provided by near-resonantimaging, the phase noise is recorded across the imaging area (dUmeas)at xRF ¼ 2MHz, as shown in Fig. 3(c). In other words, dUmeas repre-sents the total phase noise recorded while imaging in the absence of aspecimen. The optimum working point is where dUmeas is minimal(i.e., highest phase stability). Estimates of the phase noise are obtainedfrom the residual magnetic field noise after active stabilization—5 nT,regardless of the operational frequency. The intrinsic phase noise limit(dU5nT) is calculated from this value, and used for identifying the bestoperational conditions [Fig. 3(d)]. The near-resonant detuning point isdependent on the operation frequency as must be chosen accordingly.For example, at xRF ¼ 200 kHz, the detuning is þ375Hz. In the caseof Fig. 3, in resonant conditions, the minimum detectable conductivityis 160 Sm�1, as a result of the phase noise. Near-resonant operation at2MHz, with a detuning of þ750Hz, reduces the phase noise by a fac-tor 172, thus bringing the residual noise phase to dU5nT ¼ 0:03�. Thisprojects imaging to the 1 Sm�1 level, fully matching the requirementsfor imaging biological tissues2 (10 Sm�1).

Figure 4 demonstrates the tunability of the imaging system: inpanels (a)–(c), we show the 5� 102 Sm�1 sample imaged at 200 kHz,800 kHz, and 1.4MHz. Otherwise, the conditions are the same as thoseof Fig. 2. The corresponding ratios between the samples’ thickness(t¼ 10mm) and the skin depth dxRF are: 0.20, 0.40, and 0.53,

respectively [0.63 when xRF ¼ 2MHz—Fig. 2(d)]. For a penetrationdepth larger than the object’s thickness, a linear dependence of thephase on the frequency is expected,8 as anticipated

DU � xRFl0r ; (1)

where l0 is the magnetic permeability of free space. This can beobserved in the case of the low-conductivity samples, for which thiscondition is satisfied—see the lower part of Fig. 4(d). For the higherconductivities, deviations from Eq. (1) can be noticed in the upperpart of Fig. 4(d), and are attributed to the skin effect. This regulates thegeneration of eddy currents, and thus the volume contributing tothe imaging signal. At a high frequency, the skin depth for thehigh-conductivity specimens is smaller than their thickness (e.g., forthe 1� 104 Sm�1 sample, t=dxRF � 2:8 at 2MHz). Thus, the effectivevolume is reduced, and a lower increase in DUmax with frequency isobserved. The results of Fig. 4 also demonstrate that our system iscapable of resolving variations in conductivity 2, thus covering awide range of cases of interest for biomedical imaging.31

In conclusion, we have demonstrated contactless and noninvasiveelectromagnetic induction imaging of low-conductivity specimens withan ultrasensitive (50 fT/

ffiffiffiffiffiffi

Hzp

) and broadly tunable (100 kHz–2MHz)RF atomic magnetometer biased to operate in near-resonant condi-tions. Imaging was obtained in unshielded environments, nearroom temperature. Images of low-conductivity test samples—6.25 cm3

doped semiconductors measuring between 1� 104 Sm�1 and5� 101 Sm�1—represent a three orders of magnitude reduction inthe lowest-conductivity material imaged with an atomic magnetome-ter, and a substantial decrease in volume with respect to previousresults with low-conductivity samples. A large dynamic range was alsoobserved, with the largest span of conductivities ð½1� 104 Sm�1;5� 101 Sm�1�Þ simultaneously imaged, to date. Near-resonant opera-tion allowed the reduction of the phase-noise in an unshielded environ-ment to 0:03� in operational conditions. The projected performance ofour system indicates a sensitivity of 1 Sm�1 in an unshielded environ-ment, with a sensitivity to changes in conductivity smaller than a factor2, thus matching the requirements for long-sought applications in bio-medical imaging. Here, variations of conductivity between 102 and 10are expected, on a scale of several millimeters.1,2 In light of the resultspresented in this letter, this performance appears achievable in the veryshort term. Dedicated optimization and validation for specific condi-tions would then be necessary.

This work was supported by the UK Quantum TechnologyHub in Sensing and Metrology, Engineering and Physical SciencesResearch Council (EPSRC) (No. EP/M013294/1).

REFERENCES1W. T. Joines, Y. Zhang, C. Li, and R. L. Jirtle, “The measured electrical proper-ties of normal and malignant human tissues from 50 to 900 MHz,” Med. Phys.21, 547–550 (1994).

2L. Marmugi and F. Renzoni, “Optical magnetic induction tomography of theheart,” Sci. Rep. 6, 23962 (2016).

3H. Griffiths, “Magnetic induction tomography,” Meas. Sci. Technol. 12, 1126(2001).

4H. Griffiths, W. R. Stewart, and W. Gough, “Magnetic induction tomography:A measuring system for biological tissues,” Ann. New York Acad. Sci. 873,335–345 (1999).

FIG. 4. Imaging tunability. (a)–(c) Images of the 5� 102 Sm�1 sample at 200 kHz,800 kHz, and 1.4 MHz. (d) Maximum phase change across an image ðDUmaxÞ foreach of the six doped Si samples as a function of frequency. The dashed lines arethe best interpolating curves (piecewise cubic Hermite interpolating polynomials) foreach dataset.

Applied Physics Letters ARTICLE scitation.org/journal/apl

Appl. Phys. Lett. 115, 083503 (2019); doi: 10.1063/1.5116811 115, 083503-4

Published under license by AIP Publishing

5R. Merwa, K. Hollaus, O. Bir�o, and H. Scharfetter, “Detection of brain oedemausing magnetic induction tomography: A feasibility study of the likely sensitiv-ity and detectability,” Physiol. Meas. 25, 347–354 (2004).

6M. Zolgharni, H. Griffiths, and P. D. Ledger, “Frequency-difference MIT imag-ing of cerebral haemorrhage with a hemispherical coil array: numerical mod-elling,” Physiol. Meas. 31, S111–S125 (2010).

7W. Pan, Q. Yan, M. Qin, G. Jin, J. Sun, X. Ning, W. Zhuang, B. Peng, and G.Li, “Detection of cerebral hemorrhage in rabbits by time-difference magneticinductive phase shift spectroscopy,” PLoS One 10, 1–14 (2015).

8H. Griffiths, W. Gough, S. Watson, and R. J. Williams, “Residual capacitivecoupling and the measurement of permittivity in magnetic inductiontomography,” Physiol. Meas. 28, S301 (2007).

9C. Deans, L. Marmugi, S. Hussain, and F. Renzoni, “Electromagnetic inductionimaging with a radio-frequency atomic magnetometer,” Appl. Phys. Lett. 108,103503 (2016).

10A. Wickenbrock, N. Leefer, J. W. Blanchard, and D. Budker, “Eddy currentimaging with an atomic radio-frequency magnetometer,” Appl. Phys. Lett. 108,183507 (2016).

11C. Deans, L. D. Griffin, L. Marmugi, and F. Renzoni, “Machine learning basedlocalization and classification with atomic magnetometers,” Phys. Rev. Lett.120, 033204 (2018).

12P. Bevington, R. Gartman, W. Chalupczak, C. Deans, L. Marmugi, and F.Renzoni, “Non-destructive structural imaging of steelwork with atomic magne-tometers,” Appl. Phys. Lett. 113, 063503 (2018).

13P. Bevington, R. Gartman, and W. Chalupczak, “Imaging of material defects witha radio-frequency atomic magnetometer,” Rev. Sci. Instrum. 90, 013103 (2019).

14P. Bevington, R. Gartman, andW. Chalupczak, “Enhanced material defect imagingwith a radio-frequency atomic magnetometer,” J. Appl. Phys. 125, 094503 (2019).

15C. Deans, L. Marmugi, and F. Renzoni, “Through-barrier electromagneticimaging with an atomic magnetometer,” Opt. Express 25, 17911–17917 (2017).

16C. Deans, L. Marmugi, and F. Renzoni, “Active underwater detection with anarray of atomic magnetometers,” Appl. Opt. 57, 2346–2351 (2018).

17G. Chatzidrosos, A. Wickenbrock, L. Bougas, H. Zheng, O. Tretiak, Y. Yang,and D. Budker, “Eddy-current imaging with nitrogen-vacancy centers in dia-mond,” Phys. Rev. Appl. 11, 014060 (2019).

18D. Budker and M. Romalis, “Optical magnetometry,” Nat. Phys. 3, 227 (2007).19I. M. Savukov, S. J. Seltzer, M. V. Romalis, and K. L. Sauer, “Tunable atomicmagnetometer for detection of radio-frequency magnetic fields,” Phys. Rev.Lett. 95, 063004 (2005).

20M. P. Ledbetter, V. M. Acosta, S. M. Rochester, D. Budker, S. Pustelny,and V. V. Yashchuk, “Detection of radio-frequency magnetic fieldsusing nonlinear magneto-optical rotation,” Phys. Rev. A 75, 023405(2007).

21W. Chalupczak, R. M. Godun, S. Pustelny, and W. Gawlik, “Room temperaturefemtotesla radio-frequency atomic magnetometer,” Appl. Phys. Lett. 100,242401 (2012).

22C. Deans, L. Marmugi, and F. Renzoni, “Sub-picotesla widely tunable atomicmagnetometer operating at room-temperature in unshielded environments,”Rev. Sci. Instrum. 89, 083111 (2018).

23J. Belfi, G. Bevilacqua, V. Biancalana, S. Cartaleva, Y. Dancheva, and L. Moi,“Cesium coherent population trapping magnetometer for cardiosignal detec-tion in an unshielded environment,” J. Opt. Soc. Am., B 24, 2357–2362(2007).

24H. Xia, A. Ben-Amar Baranga, D. Hoffman, and M. V. Romalis,“Magnetoencephalography with an atomic magnetometer,” Appl. Phys. Lett.89, 211104 (2006).

25K. Jensen, M. A. Skarsfeldt, H. Stærkind, J. Arnbak, M. V. Balabas, S.-P.Olesen, B. H. Bentzen, and E. S. Polzik, “Magnetocardiography on an isolatedanimal heart with a room-temperature optically pumped magnetometer,” Sci.Rep. 8, 16218 (2018).

26J. F. Barry, M. J. Turner, J. M. Schloss, D. R. Glenn, Y. Song, M. D. Lukin, H.Park, and R. L. Walsworth, “Optical magnetic detection of single-neuron actionpotentials using quantum defects in diamond,” Proc. Natl. Acad. Sci. 113,14133–14138 (2016).

27A. Vander Vorst, A. Rosen, and Y. Kotsuka, RF/Microwave Interaction withBiological Tissues (John Wiley & Sons, 2006), Vol. 181.

28M. Lu, W. Zhu, L. Yin, A. Peyton, and W. Yin, “Reducing the lift-off effect onpermeability measurement for magnetic plates from multi-frequency inductiondata,” IEEE Trans. Instrum. Meas. 67(1), 167 (2017).

29M. O’Toole, N. Karimian, and A. J. Peyton, “Classification of non-ferrousmetals using magnetic induction spectroscopy,” IEEE Trans. Ind. Inf. PP, 1(2018).

30G. Bevilacqua, V. Biancalana, Y. Dancheva, and A. Vigilante, “Self-adaptiveloop for external-disturbance reduction in a differential measurement setup,”Phys. Rev. Appl. 11, 014029 (2019).

31C. Deans, L. Marmugi, S. Hussain, and F. Renzoni, “Optical atomic magnetom-etry for magnetic induction tomography of the heart,” Proc. SPIE 9900, 10(2016).

Applied Physics Letters ARTICLE scitation.org/journal/apl

Appl. Phys. Lett. 115, 083503 (2019); doi: 10.1063/1.5116811 115, 083503-5

Published under license by AIP Publishing